Contamination Level Estimation Method for High Voltage Insulators

1. A contamination level estimation method for high voltage insulators, comprising the steps of: building an image data set of a collection of naturally polluted insulator samples; estimating equivalent salt deposit density (ESDD) levels of the collection, the ESDD estimations being used for training a Multi-layer Feed-forward Neural Network (MFNN); forming a grayscale model from an RGB representation of each of the images; a first stage segmentation step comprised of segmenting the grayscale insulator images to exclude an insulator cap from the image using matrix manipulation; a second stage edge based segmentation step comprised of segmenting the grayscale insulator images to exclude the background; transforming the RGB representation of each of the images to an HSV representation of each of the images; extracting statistical and algebraic features from the HSV representation of each of the images, wherein the HSV segmented images are used to assess the contamination level that would lead to a flashover; using the statistical and algebraic features and the ESDD estimates to train the (MFNN) to correlate levels of contamination captured in insulator images with ESDD levels, the contamination levels being light pollution level, medium pollution level, heavy pollution level and very heavy pollution level categories; classifying live insulator images using the MFNN; and outputting to a user the classified high voltage insulator pollution levels as a corollary to equivalent salt deposit density (ESDD).

4. The contamination level estimation method for high voltage insulators according to claim 1 , wherein the MFNN training step further comprises the step of using a back-propagation algorithm to find optimum weight biases of the MFNN along a negative gradient of a cost function, the cost function being characterized by the relation, J = ∑ i = 1 n ⁢ [ u ⁡ ( i ) - u r ⁡ ( i ) ] , where u and u r denote actual and desired outputs respectively.

6. A computer software product, comprising a non-transitory medium readable by a processor, the non-transitory medium having stored thereon a set of instructions for performing a contamination level estimation method for high voltage insulators, the set of instructions including: (a) a first sequence of instructions which, when executed by the processor, causes said processor to build an image data set of a collection of naturally polluted insulator samples; (b) a second sequence of instructions which, when executed by the processor, causes said processor to estimate equivalent salt deposit density (ESDD) levels of the collection, the ESDD estimations being used for training a Multi-layer Feed-forward Neural Network (MFNN); (c) a third sequence of instructions which, when executed by the processor, causes said processor to form a grayscale model from an RGB representation of each of the images; (d) a fourth sequence of instructions which, when executed by the processor, causes said processor to perform a first stage segmentation step comprised of segmenting the grayscale insulator images to exclude an insulator cap from the image using matrix manipulation; (e) a fifth sequence of instructions which, when executed by the processor, causes said processor to perform a second stage edge based segmentation step comprised of segmenting the grayscale insulator images to exclude the background; (f) a sixth sequence of instructions which, when executed by the processor, causes said processor to transform the RGB representation of each of the images to an HSV representation of each of the images; (g) a seventh sequence of instructions which, when executed by the processor, causes said processor to extract statistical and algebraic features from the HSV representation of each of the images, wherein the HSV segmented images are used to assess the contamination level that would lead to a flashover; (h) an eighth sequence of instructions which, when executed by the processor, causes said processor to use the statistical and algebraic features and ESDD estimates to train the (MFNN) to correlate levels of contamination captured in insulator images with ESDD levels, the contamination levels being light pollution level, medium pollution level, heavy pollution level and very heavy pollution level categories; (i) a ninth sequence of instructions which, when executed by the processor, causes said processor to classify live insulator images using the MFNN; and (j) a tenth sequence of instructions which, when executed by the processor, causes said processor to output to a user the classified high voltage insulator pollution levels as a corollary to equivalent salt deposit density (ESDD).

9. The computer software product according to claim 6 , further comprising: seventeenth sequence of instructions which, when executed by the processor, causes said processor to use a back-propagation algorithm to find optimum eight biases of the MFNN along a negative gradient of a cost function, the cost function being characterized by the relation, J = ∑ i = 1 n ⁢ [ u ⁡ ( i ) - u r ⁡ ( i ) ] , where u and u r denote actual and desired outputs respectively.